BEFA report: The built environment sector needs responsible AI adoption
Posted on: 21 July, 2026
Artificial intelligence is already being used across the built environment, but the sector needs to move from fragmented experimentation towards responsible, professional and accountable adoption, finds the new ‘Mapping the AI Landscape in the Built Environment’ report from the University’s Built Environment Futures Assembly (BEFA).
Produced with Harlow Consulting, the report examines how AI is currently being used across design, engineering, planning, construction and related professional or technical disciplines, as well as what this means for future workforce development.
The report is based on a workshop with BEFA’s Strategic Advisory Board, 13 in-depth interviews with senior sector figures from industry, professional bodies, academia and government, and a desk review of current initiatives, standards, guidance and examples of real-world AI adoption.
From 157 minutes to 37 minutes
The report points to early evidence of AI’s potential to reduce time spent on routine tasks. Research cited in the report found that risk assessments were reduced from 157 minutes to 37 minutes, 10,000-word reports could be summarised by MS Copilot in 30 minutes, and bid preparation was reduced from three weeks to three days.
However, adoption remains uneven, often informal and frequently ahead of the governance frameworks needed to make its use consistent and valuable.
The greatest near-term risk identified by the report is unmanaged use, rather than mass job loss. AI outputs can be fast, fluent and plausible, which makes them useful, but also risky when they are not properly checked.
Mark Farmer, Chair of the Built Environment Futures Assembly, said: “AI may support the heavy lifting of some tasks, but responsibility and accountability must remain with competent professionals. The built environment does not need to chase every emerging tool. It needs the competence, confidence and governance required to use AI well, and to ensure adoption is shaped by the needs of the sector rather than by technology alone.”
From experimentation to responsible adoption
A central theme of the report is that the built environment’s AI challenge is whether the sector can integrate it responsibly into professional workflows.
The report argues that AI adoption should be understood as a workforce, governance and professional practice issue, rather than a technology or IT issue.
This matters because the built environment is a safety-critical and highly accountable sector. AI tools can help improve productivity, support better decision-making and reduce routine administrative burden, but unmanaged adoption could also accelerate weak processes, magnify poor data and introduce unverified material into decisions that affect safety, cost, compliance or client confidence.
AI as augmentation, not autonomy
The report’s core message is that AI should augment professional expertise, not bypass it. The value of built environment professionals will increasingly lie in framing the right questions, selecting appropriate tools, understanding data quality, challenging outputs and making safe, relevant and defensible decisions.
In practice, this means professionals need to treat AI as support for decision-making as opposed to a substitute for judgement. The report concludes that AI can support professional work, but it cannot (and should not) carry professional responsibility.
This people-first approach also requires the sector to be clearer about what it wants AI to do. The report warns against simply reacting to vendor-led adoption or using AI to make existing inefficient processes marginally faster.
Verification, data and assurance
Data quality is another major theme. The report warns that AI will expose the sector’s long-standing data problems, including fragmented information, inconsistent standards, weak interoperability and unclear data ownership. Data literacy, information quality, provenance, BIM, common data environments and handover information are therefore becoming mainstream professional competencies, rather than specialist digital concerns.
The report also highlights the role of clients, regulators, insurers and procurement teams in shaping responsible adoption. They will increasingly need to understand not just whether AI is being used, but how it is being used, with what data, under what controls and with what human review.
Skills, competence and early-career pathways
A key finding is that AI literacy is becoming a mainstream professional competency, not a specialist digital skill. The report cites evidence that only 30% of built environment firms provide AI literacy training, despite the increasing use of AI for drafting, searching, coding, summarising and analysis. It argues that professionals will need to understand not only how to prompt AI tools, but how to select the right tool, assess the quality of underlying data, challenge outputs and know when human judgement or deterministic methods are required.
The report also raises a particular concern about early-career pathways. Many of the tasks AI can most readily accelerate, such as drafting, basic analysis, data handling and routine problem-solving, are also the tasks through which junior professionals traditionally build technical understanding, judgement and confidence.
If these learning opportunities are reduced or bypassed without being replaced by new, structured development pathways, the sector risks weakening its future talent pipeline. Employers, educators and professional bodies will therefore need to redesign apprenticeships, graduate schemes, vocational learning, professional formation and CPD so that AI supports learning rather than bypassing it.
Key conclusions
The report sets out a series of recommendations for industry, education providers, professional bodies, clients and policymakers. These include the need for a clearer shared direction on responsible AI, stronger verification practices, better data and information capability, practical competency frameworks, redesigned early-career learning and more role-specific training.
Read the full report: ‘Mapping the AI Landscape in the Built Environment’
A summary version of the report, pulling out the key findings, is available here
